article relationships material properties

8/10/2019 Article Relationships Material Properties

1/24

Mechanical performance of polymer systems: The relation

between structure and properties

Han E.H. Meijer*, Leon E. Govaert

Materials Technology, Dutch Polymer Institute, Eindhoven University of Technology, W-hoog 4.140, P.O. Box 513,

5612 MB Eindhoven, The Netherlands

Accepted 24 March 2005

Abstract

A direct relation between molecular characteristics and macroscopic mechanical properties of polymeric materials was

subject of a vast number academic and industrial research studies in the past. Motivation was that an answer to this question

could, in the end, result in guidelines how to construct tailored materials, either on the molecular level or, in heterogeneous

materials, on the micro-scale, that could serve our needs of improved materials without the need of extensive trial and error

work. Despite all attempts, no real universally applicable success was reported, and it was only after the introduction of the

concept of the polymers intrinsic deformation behavior that some remarkable progress could be recognized. Thus, it is first

important to understand where intrinsic deformation behavior of polymeric materials stands for. Second, it is interesting to

understand why this intermediate step is relevant and how it relates to the molecular structure of polymers. Third and, in the end,the most computational-modeling-based question to be answered is how intrinsic behavior relates to the macroscopic response

of polymeric materials. This is a multi-scale problem like encountered in many of our present research problems.

q 2005 Published by Elsevier Ltd.

Keywords: Amorphous polymers; Toughness; Intrinsic behavior; Mechanical properties; Constitutive modeling; Multi-level finite element

method; Structureproperty relations

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 916

2. Structure of polymers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 917

3. Intrinsic deformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 919

4. Physical ageing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 922

5. From intrinsic behavior to macroscopic response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 924

6. Life time prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 925

Prog. Polym. Sci. 30 (2005) 915938

www.elsevier.com/locate/ppolysci

0079-6700/$ - see front matter q 2005 Published by Elsevier Ltd.

doi:10.1016/j.progpolymsci.2005.06.009

*Corresponding author.

E-mail address:[email protected] (H.E.H. Meijer).

URL: www.mate.tue.nl.
http://www.elsevier.com/locate/ppolyscihttp://www.elsevier.com/locate/ppolysci


2/24

7. Craze initiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 927

8. Heterogeneous systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 930

9. Optimal toughness modifier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 931

10. Brittle-to-tough transitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 933

11. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 934

Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 935

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 935

1. Introduction

Bridging the gap between molecular and macro-

scopic properties, with as extra intermediate step the

processing history, seeFig. 1a, is a just as challenging

as impossible task. Reason is that the multi-scale

problems involved require far too much compu-tational time and computational memory to be

resolved via ab initio analyses. Analyses on different

scales require averaging the resulting properties from

the analysis on the underlying scale, consequently

loosing detail. Therefore, alternatives are searched

f or . A big s te p f or wa rd was made wit h t he

development of a video-controlled tensile test by

Christian GSell at the Ecole des Mines in Nancy

[1,2]. This technique enabled the experimental

assessment of the intrinsic deformation behavior: the

polymers stress response measured during homo-

geneous deformation. More or less simultaneously,Mary Boyce, at MIT, optimized the much more

practical uniaxial and plain strain compression tests to

obtain similar information in different loading

geometries [3,4]. Reason to be pleased with this

intermediate step is twofold. First, our present

Fig. 1. From molecular details to macroscopic response via the intermediate of the polymers intrinsic deformation.

H.E.H. Meijer, L.E. Govaert / Prog. Polym. Sci. 30 (2005) 915938916


3/24

advanced computational tools and expertise can be

successfully used in the relation between the intrinsic

behavior and the macroscopic response, the right-

hand side ofFig. 1b. Prerequisite for the success ofthis route was the development of constitutive models

that could capture the intrinsic behavior. So let start to

focus on this issue. It is known that it took the polymer

melt rheology community some half a century to

arrive at constitutive equations that are not only valid

in simple shear flows but also can deal with

extensional flows. The (extended) Pom Pom model

quantitatively describes the melt rheology for

branched, and even also for linear, polymers, in

complex flows and is based on molecular character-

istics [57]. Since rheology is based on fluidmechanics, rheologists can be considered as succes-

sors of Isaac Newton, thus mathematically educated

scientists. Therefore, it is at least remarkable that it

took them that much time. The main topics of the

rheologists concern were the quantitative modeling

of the deformation rate dependent shear and elonga-

tional viscosity and first (and eventually second)

normal stress difference, in incompressible homo-

geneous time dependent start-up flows. Interesting

now is that the major problem encountered in the solid

state rheology of the same polymers is that they are

per definition tested in compressible inhomogeneous

start-up flows in extension. Solid state rheologists can

be considered as the successors of Robert Hooke.

They are practical engineers that have to deal with the

impossible interpretation of the most widely used test:

the tensile test on a normalized dogbone-shaped

sample that, at best, shows only a simple localization

behavior, like necking. All relatively recent progress

reported in solid state rheology made of course use of

the results obtained in the constitutive equations for

melts and the good news is that at present the

macroscopic response of amorphous polymers to

shear and tensile loading can now be quantitatively

described. In plain strain 2D, or even in completely

3D computations, in short and in long time loading, in

homogeneous and also in heterogeneous systems,

once the polymers intrinsic behavior (and its

microstructure in case of heterogeneous systems) is

known. Use is made of standard (e.g. Marc, Abaqus)

software combined with custom made (Multi-Level-

Finite-Element) analyses.

Second,the relation between molecular properties

of polymers and their intrinsic behavior, the left-hand

side of Fig. 1b, are accessible via well defined

experiments that are much easier interpretable thanwhen compared to the results of tensile tests, creep

and fatigue tests or Izod impact tests. So let us focus

now on the structure of polymers.

2. Structure of polymers

Polymers are different from other construction

materials like ceramics and metals, because of their

macromolecular nature. The covalently bonded, long

chain structure makes them macromolecules anddetermines, via the weight averaged molecular

weight, Mw, their processability, like spin-, blow-,

deepdraw-, generally melt-formability. The number

averaged molecular weight, Mn, determines the

mechanical strength, and high molecular weights are

beneficial for properties like strain-to-break, impact

resistance, wear, etc. Thus, natural limits are met,

since too high molecular weights yield too high shear

and elongational viscosities that make polymers

inprocessable. Prime examples are the very useful

poly-tetra-fluor-ethylenes, PTFEs [8,9], and ultra-

high-molecular-weight-poly-ethylenes, UHMWPEs

[10], and not only garbage bags are made of poly-

ethylene, PE, but also high-performance fibers

[1113] that are even used for bullet proof vests

(alternatively made from, also inprocessable in the

melt, rigid aromatic polyamides [14,15]). The

resulting mechanical properties of these high per-

formance fibers, with moduli of 150 GPa and

strengths of up to 4 GPa, represent the optimal use

of what the potential of the molecular structure of

polymers yields, combined with their low density.

Thinking about polymers, it becomes clear why livingnature used the polymeric concept to build its

structures. And not only in high strength applications

like wood, silk or spider-webs.1

1 About natures processing: What is the melting point of

ceramics? And that of metals? If these materials are to be processed

at ambient temperature and atmospheric pressure, then what is their

solubility? How does nature use chemistry to make teeth and bones?

Most of this is known and, interestingly, comes rather close to

techniques we use in polymer technology.

H.E.H. Meijer, L.E. Govaert / Prog. Polym. Sci. 30 (2005) 915938 917


4/24

Let us now consider diamond, a covalently bonded

structure of carbon atoms. What is the energy to create

new surfaces in diamond, known as the strongest

material in nature? The answer is 1 J/m2. How much is

that? It is the energy that comes free if an ordinary

apple falls from an ordinary table [16]. And that is

shocking, especially because in this experiment we

splitted the two surfaces at once, without applying any

zipping process. It is the interaction potential that

quantifies the energy needed to split atoms, and thus

also to create ideal surfaces made of those atoms. Howthus this explain the impact toughness of a standard,

commodity polymer like PE, which is chemically

spoken diamond with two out of the four strongest

bonds in nature (covalent) replaced by two of the

weakest bonds (Van der Waals), and that measures, in

a standard Izod impact test, a value of 105 J/m2? The

answer to this question is delocalization. If a material

is able to delocalize its strain, out of the area of the

largest stress, it will undergo deformation in the

volume. This volume deformation is, in a standar-

dized test like Izod-impact, still attributed to thegeometrical cross-sectional area in the test bar, behind

the notch, seeFig. 2.2

In the glassy state the polymers secondary bonds

determine the elastic part of the mechanical properties,

modulus and yield stress, while the primary bonds in the

network potentially induce the necessary delocalizationof local strain that gives toughness. Given this

mechanism we conclude that all amorphous polymers

with a Tg above room temperature have roughly the

same elastic Youngs (not Hookes3) modulus of 3 GPa.

Low, compared to both competitors, ceramics and

metals, since controlled by their weak secondary bonds.

Even worse becomes the case for semicrystalline

polymers with a Tg below room temperature, that

should be considered as a gel of an amorphous fluid

physically kept together by higher melting crystals.

Only in high performance fibers, like UMWPE andaromatic polyamides, the covalently bonded structure is

mechanically addressed during loading, provided that

sufficient length is given to the weak secondary

interactions to transfer the load from one molecule to

the other. This, of course, only works in the direction of

the processing-induced 100% main chain orientation.

Non-oriented polymers basically only strongly differ in

impact strength or, equivalently, their ability to

delocalize strain. Ceramics are intrinsically brittle but

some are less brittle, like man-made ceramic fiber

reinforced ceramics[17]or the natural multi-layeredtoothor seashell ceramics [18,19], the last with typically

400 nm sized layered structures. Delocalization in the

artifacts is realized because the ceramic fibers survive

the breaking and bridge the cracks, giving rise to both a

second dissipation mechanism via debonding along the

fibers andin the end a possibly local strainhardening via

cooperative fiber loading. This is also the way stone

reinforcement in concrete works. Delocalization in the

natural objects is induced by weakening the structure

perpendicular to the loading direction yielding a

multiple cracking mechanism, see Fig. 3. This principle

is comparable to the change of diamond into

Fig. 2. Diamond, PE and the notched impact test.

2 We should define impact values only in J/m2. Either or not

correct, volume deformation is attributed to the surface energy. A

value of 100.000 J/m2 for PE is of course rather arbitrary. It

represents the maximum energy absorbed in the (optically stress-

whithening) volume during polymer testing using the Izod test-bar,

and it is partly determined by the sample geometry. A value like that

for PE means in practice: no-break.

3 The power of any springy body is in the same proportion

with the extension (Sic vis, sic extensio) was Hookes way in

1676 to express the constitutive equation for a solid that we now

write as sZGg, or sZG B in 3D, where Newtons 1687

definition of the viscosity for fluids was The resistance which

arises from the lack of slipperiness of the parts of the liquid,

other things being equal, is proportional to the velocity with

which the parts of the liquid are separated from one another,

that we now write as sZhg_, or sZ2h D in 3D.



5/24

polyethylene, weakening the structure in some direc-

tions, thus suffering in mechanical bulk behavior like

modulus and yield strength and, where seashells are still

pretty hard and strong, polyethylene will never be

diamond anymore.

3. Intrinsic deformation

Our basic question was: where does the intrinsicbehavior of polymers originate from and can we

understand it? To answer that, we start with returning

to the original work of Haward and Thackray of 1965

who described the materials response with two

parallel processes: (i) the initial non-linear elastic

response up to yield, controlled by the secondary,

intermolecular, interactions, combined with (ii) the

entanglement network response in parallel from the

primary intramolecular interactions which gives

an entropic contribution at large strains [20]. In

Hawards approach, see Fig. 4, the secondary

interactions were modeled via a Maxwell model

(fluid!), where all non-linearity was put into the stress-

activated viscosity function that followed an Eyring

model. The original model not only received a typicalpolymer type of response in terms of citations with a

change in slope from 1 to 3 at the onset of the use of

continuum mechanics based computer modeling[21],

but also basically was at the cradle of the solid state

rheology. Its conceptual approach was followed and

Fig. 3. Breaking of homogeneous (left) and heterogeneous (right) ceramics. The figure on the left shows catastrophic strain localization in a

Californian Giant Rock induced by thermal stresses from a nightly fire. The figure on the right shows delocalized strain in seashell ceramics, see

Refs.[18,19].

total

inter-molecular

network

true strain

truestre

ss

100 1010

5

10

15

20

25(b)(a)

eq[MPa]

log()[MP

as

]

Fig. 4. The stress decomposition proposed by Haward and Thackray [20](a). For its present 3D formulation, seeTable 1.Further we show the

characteristic stress-dependence of the viscosity of the dashpot for PC at room temperature [27](b).



6/24

extended to full 3D constitutive relations by several

research groups, e.g. those of Mary Boyce at MIT[3,22,23], Paul Buckley in Oxford [2426] and our

group at the TU/e in Eindhoven [2729]. Common

factors in these models are the application of rubber

elasticity to model strain hardening and a stress-

dependent viscosity to capture the deformation

kinetics. Similar to temperature, an applied stress

leads to an increase in segmental mobility [30,31]

with a resulting decrease in viscosity; in extreme this

leads to yielding, the onset of flow, induced by stress

(stress-induced glass transition). In the case of

polycarbonate, PC [27] an equivalent stress of30 MPa lowers the viscosity of the material with 12

decades (seeFig. 4b)!

Knowing that for atactic polypropylene, PP,4 ab

initio calculations by Doros Theodorou and Uli Suter,

at that time at MIT, showed that affine deformations

along the chain do not exists outside a region of

roughly 10 C-atoms[32], we can still today conclude

that the process of plastic deformation in amorphous

polymers is controlled by molecular motions on a

segmental scale. Thus molecular weight and entangle-

ments should not play a dominant role. Only thesecondary interactions and the free volume kinetics

could be of primary interest. Once the critical stress is

surpassed, the stress to further flow should be

constant, until the entanglement network starts to

orient, manifesting itself in strain hardening

(represented by the parallel spring in the Haward

model). Here, it is the entanglements that control themodulus and it is the molecular weight that controls

the strength at breakage. In this picture there seems to

be no reasoning for strain softening. This now is

somewhat more important than its first glance seems

to be, seeFig. 5.

The combination of strain softening, after yield,

followed by strain hardening, at larger deformations,

proves to be the key issue in understanding mechanical

behavior. Their mutual ratio determines whether a

polymer will manifest non-catastrophic localization

(like in bullet-proof polycarbonate PC), or will show

catastrophic localization in terms of craze formation

perpendicular to the loading direction (like in the brittle

polystyrene PS) [33]. The origin of strain softening thus

seems to be important and it was only after reaching an

extreme intrinsic behavior that some progress was

realized. What is extreme behavior? Basically, it is

the response of a completely fresh material. Best way to

realize that state is by mechanical rejuvenation, rather

than by applying thermal rejuvenation5 (that are not

identical [34]). This is the best illustrated by the fact that

the lowest yield stress reached via thermal rejuvenation

in PC is approximately 57 MPa, while mechanicalrejuvenation of PC at sufficiently high strains (gZ0.3)

yields 35 MPa.6 In PC, mechanical rejuvenation can be

0 0.5 1 1.50

50

100

150

PC

PS

true strain []

truestress[M

Pa]

1 1.05 1.1 1.150

25

50

75(b)(a)

[]

eng.stress[M

Pa] PS, rolled

PC, rolled

Fig. 5. Intrinsic behavior of polycarbonate PC and polystyrene PS before (a) and after (b) mechanical rejuvenation.

4 The only polymer that at that time could properly be modeled

since it did not suffer from either large geometrical hindrance (like

in polystyrene PS) or from polar interactions (like in poly-

vinylchloride PVC), neither did it crystallize.

5 Melting followed by fast quenching to the solid state is referred

to as thermal rejuvenation.6 Discussions on whether mechanical rejuvenation is the same as

thermal rejuvenation still continue. The answer is simple: It is not,

because mechanical rejuvenation is superior, since during the

quench in thermal rejuvenation, the polymer starts to considerably

age already.



7/24

achievedby twistinga cylindrical test bar to andfroover

7208at room temperature prior to tensile testing[35].

For lower entanglement density polymers like PS this

does not work (crazes occur) and an alternative routewas explored in deforming the material at room

temperature under (compressive) stresses via a 30%

height reduction in a two-roll milling process[36,37].

For PC, 10% proves to be sufficient to induce

homogeneous deformation without neck formation.

Also, PS shows homogeneous deformation, without

neck, but also without craze formation, seeFig. 6.

Our understanding is that once the secondary bonds

are locally weakened by stress, the material must flow

under a constant stress up to the point that strain

hardening sets in. And that is exactly what we find inrejuvenated materials. During the cold rolling process

we pushed the original, aged, material to a strain that

was sufficiently large that we entered the start the strain

hardening part of the intrinsic deformation curve. By

the way, interestingly enough, the density of

the polymers increased during this process of mechan-

ical rejuvenation. After release of the stress and

reloading, the yield stress proved to be reduced to theextent that no softening occurs anymore. The cause of

strain localization is thus removed and that explains the

no-necking for PC and the no-crazing of PS. The low

yield stress, though, does not remain and in time, ageing

time, it increases, leading to a return of softening and

thus thereturn of brittleness. ForPS (with its lowdensity

thus high free volume, low entanglement density, high

chain stiffness and thus high molecular weight between

entanglements of about 20.000), brittleness returns

within hours after rejuvenation and for PC (Mez2000)

it is a matter of monthsbefore necking is observed again,seeFig. 7.

In classical free volume theories for ageing, an

increase in density is always accompanied with an

increase in yield stress. During mechanical

Fig. 6. Deformation in the quenched (a and c) and mechanically rejuvenated state (b and d) for PC (a and b) and PS (c and d) [33].



8/24

rejuvenation the opposite occurred. Physical ageing

experts tried to make the free volume theory the holey

Grail of polymer physics. They failed but later

explained that they, of course, meant the free volume

distribution being the responsible driving force[38].

We conclude from our experiments that free volume is

just a, let us say, related property.

4. Physical ageing

What causes the yield stress to increase in time

upon ageing? Already in the 1970s it was quantified

that upon ageing of amorphous polymers in calori-

metry, DSC, a pronounced enthalpy peak develops

near the glass transition temperature [39,40]. Simul-

taneous with the development of this enthalpy

overshoot, the yield stress is observed to increase

[41,42]. Both effects can be rationalized in terms of an

evolution of the potential energy landscape[4346].The change in time of the energy landscape, that is

the local attraction of individual atoms to their

neighbours is illustrated in Fig. 8c and shows a

local densification upon physical ageing. Indeed, a

free volumedistributionif you want, but rather a local

attraction-energy distribution. In order to induce

molecular mobility, this local order has to be

destroyed. And that is reflected in the growing peak

in the DSC signal, Fig. 8b. If segmental mobility is

increased not by temperature, but by stress, than we

find a growing peak in stress in the stressstrain curve,

Fig. 8a, that weinterpretas an increase in yield stress

followed by softening.

Nevertheless, this last combination introduceslocalization of strain that can be catastrophic when

the ratio softening/hardening becomes overcritical

and a tough-to-brittle transition is found.

Once in macroscopic homogeneous deformation,

e.g. during mechanical rejuvenation via cold rolling,

the minima in the energy landscape are pulled out by

stress, the energy landscape is flattened, the yield

stress lowered, the material is made fresh again and

does not show strain softening and, consequently,

deforms homogeneously.7

Within the constitutive model the changes in the

yield stress, as a result of physical aging or

mechanical rejuvenation, are captured in the evolution

of a single state parameter:S. The value of this state

parameter indicates how strongly the present situation

of the polymer deviates from the reference state, in

our case the fresh, mechanically rejuvenated state.

Plotting the yield stress versus the logarithm of strain

rate, an increase of S basically shifts the curves

horizontally towards lower strain rates, see Fig. 9a.

At a constant strain rate this results in an increase of the

yield stress compared to that of the rejuvenated state.

To capture the evolution ofSas a function of timeand deformation, it is assumed to consist of two

contributions that act independently:SZSaRg. One is

the age of the sample,Sa. The other is the contribution

of mechanical rejuvenation, Rg,8 that decreases the

yield stress with ongoing total applied plastic shear.

As a result of plastic deformation the yield stress

characteristic shifts back to the reference (rejuve-

nated) state. In a constant strain rate experiment this

leads to a drop of the yield stress with increasing strain

and the intrinsic stressstrain curve evolves gradually

to that of the rejuvenated state.Fig. 10a compares the intrinsic response of an

annealed and a quenched PC sample. Annealing

results in an increase of both modulus and yield stress,

102 104 10640

50

60

70

PC

PS

1 hour 1 day 1 week

ageing time [s]

yieldstress[MPa

]

Fig. 7. Increase in yield stress in time upon physical ageing aftermechanical rejuvenation.

7 Provided that the classical Considere criterion of 1885 is not

met, implying that ratio of yield stress and strain hardening modulus

should be smaller than 3 to obtain homogeneous deformation also in

the absence of strain softening [33].8 Basically a function with an initial value of RgZ1, that

decreases towards zero with increasing applied plastic strain.



9/24

but upon plastic deformation the differences between

both curves disappear and they fully coincide at a

strain of approximately 0.3. All influence of thermal

history is erased at that strain and both samples are

transformed to the same mechanically rejuvenated

state.

From Fig. 10a it is clear that an increase of

yield stress, due to a thermal treatment, will

directly imply an increase in strain softening. The

influence of molecular weight on the intrinsic

response is usually small and negligible [29,47],

which makes thermal history the key factor in

influencing the intrinsic properties of a specific

polymer glass.

The influence of progressive physical aging is

included through a time-dependent evolution of the

true strain

truestress

ageing(a) (b)

(c)

temperature

Cp,n

ageing

distance

energylandscape

ageing

Fig. 8. Increase in yield stress (a) and DSC signal (b) in time upon physical ageing after mechanical rejuvenation and a sketch of the change in

energy landscape (c).

log(strain rate)

(a) (b)

yieldstress aged

rejuvenatedSa

R

true strain

truestress

aged

rejuvenated

Sa

Fig. 9. Schematic representation of the influence of physical aging and strain softening on the strain rate dependence of the yield stress (a).

Schematic representation of the intrinsic stressstrain curve indicating the influence of physical aging and strain softening (b).



10/24


11/24

the neck reaches the clamps, the stress starts toincrease until breakage occurs in the thinnest section

of the bar.

Performing the same analysis on PS, seeFig. 14, we

directly recognize that the strain hardening is not

sufficient to delocalize the region of the neck. This

remains the case even if we do not stop the intrinsic

behavior curve but continue to infinity, leading also to

infinite stresses within the fibril formed. Even then the

deformation remains localized, since the force

(the product of the stress times the fibrils cross-section)

that the fibril can exert on the remainder of the test bar isinsufficient. The stress can go to infinite, the cross-

section goes faster to zero. The conclusion is that PS is

basically too easy to flow, too ductile to be tough

(knowing that toughness is equivalent to delocalization

of strain). We now can understand that the arrow in

Fig. 1b on the right-hand side can only run in one

direction: from intrinsic to macroscopic and it illustratesthe difficulty of interpreting macroscopic tests.

6. Life time prediction

With the single state parameter SZSaRg we were

able to predict the polymers situation from the cradle

to the grave. In order to demonstrate this, both the

long-term behavior of polymers during static and

dynamic loading is investigated as well as the

influence of the processing conditions during theformation of the polymer on the development ofSand

thus on the yield stress. Before starting the long-term

loading experiments, the polymers age is first

determined by a single tensile test to find its yield

stress which was converted into the present value of

S. Next, long-term loading tests are performed under

Table 1

A single-mode 3D, non-isothermal constitutive equation for polymer solids, for explanation and details please download the thesis [29]

sZssCsr ht; T;p; SZh0;rT t=t0

sinht=t0exp mp

to

expSt; gr

ssZKJK1ICG~Bde ho;rTZh0;r;ref$exp

DUyR

1TK

1Tref

h i; t0Z

kTVy

srZGr~Bd Stefft; T; t; gpZSatefft; T; t$Rggp

_JZJtrD RggpZ1Cr0$expgpr1

r2K1

r1 =1Cr0r1

r2K1

r1

~Bo

eZDdKDp~BeC ~BeD

dKDp

Satefft; T; tZc0Cc1$logtefft; T; tCta

DpZ s

ds

2ht;T;p;S teffZt0

aK1T Tt0aK1s tt

0dt0

tZ

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi12trsds $s

ds

q aTTZexp

DUaR

1TK

1Tref

h i

gpZffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2$trDp$Dp

p astZ

t=tasinht=ta

; taZkTVa

0 1 2 3 40

50

100

150

200(a) (b)

80/20

60/40

20/80

40/60

2- 1/[]

truestress[MP

a]

1 1.05 1.1 1.15 1.2 1.250

25

50

75

100

80/20

60/40

20/8040/60

[]

eng.stress[MP

a]

Fig. 11. Norylw PSPPE blends in compression (a), and tensile loading (b).



12/24

different values of the (constant) stress applied, until

creep rupture occurs; see the symbols inFig. 15a.

Then simulations are performed, the line in

Fig. 15a. Because stress-enhanced physical ageing

during the tests is incorporated in the simulations,

also the endurance limit can be predicted [49].

Remarkable is the accuracy of the predictions that

did not use any adjustable parameter. Similar

comments can be made on the fatigue loading

results, see Fig. 15b.

That was the grave, so how about the cradle? Using

a simple rectangular mould and a mould-filling and

cooling analysis (Moldflow), the place and thickness

dependent thermal history can be computed every-

where in the mould, seeFig. 16a. Subsequently, it is

assumed that the evolution of yield stress, during

cooling from the melt to belowTg, is governed by the

same evolution kinetics of the state parameter S, as

observed during progressive aging, Fig. 10b. This

enables us to translate the (local) thermal histories to

(local) values of the state parameterSand, with that,

predict the distribution of yield stress over the sample

[50], see Fig. 16b. As shown in Fig. 16c, the

distribution of yield stress depends strongly on the

mould temperature, that apart from the sample

thickness is the key parameter determining the local

cooling rate.

In Fig. 16d, the predictions are compared with

experiments using different mould temperatures. The

computed average yield stresses compare amazingly

well with the experimentally determined values, the

symbols inFig. 16b. An extension of this results by

modeling polymers other than PC, such as e.g. PS,

PVC and later HDPE, is of our present research

Fig. 12. Tensile testing of PC. After neck formation at yield strain softening sets in, the stress in the remaining part of the test bar decreasessubstantially due to unloading because of strain localization in the neck. Subsequently, strain hardening occurs in he neck, accompanied by an

increase in stress and the neck starts to extend to the whole test bar.



13/24

interest. Motivation is that this results clearly opens

the possibility to predict a polymers life time

performance without performing a single mechanical

test.

7. Craze initiation

Although interesting so far, the polymer story is

not finished yet with the results obtained on the

intrinsic behavior that can reversibly at wish be

altered by physical ageing and mechanical rejuve-

nation. Neither are we finished after the quantifi-

cation of the polymers response, using advanced

constitutive equations that capture the physicsappropriately and allow us to predict the polymers

future in dependence of its past given the transient

stress and temperature profiles it will experience

during use. First reason to continue the analysis is

that if we scratch, e.g. bullet proof PC, upon

impacting the material breaks brittle via crazing.

Moreover, we also want to improve on brittle

polymers like PS, to prevent them from crazing and

introduce shear yielding. Crazing has been studied

in great detail by a large number of experts from

which Ed Kramer in Cornell, Itaca (now in Santa

Barbara) was, without any doubt, the most original

one.10 He designed beautiful experiments and

developed now well-established theories [5159].

His general starting picture is that, after plasticflow, a craze develops via the initiation of a cavity,

the growth of instabilities and finally

the coalescence of holes. In order to tackle the

problem here in a simple way, we will focus on

the initiation of the first hole in the material, underneath

a scratch or a surface defect. For that we need a craze-

initiation criterion and to identify that we performed

micro-indentation tests with a sphere of 250mm pushed

into the surface of quenched and annealed PS samples

using different normal loads [60]. After the experiments

Fig. 13. Tensile testing of PC. From intrinsic to macroscopic behavior. We plot the true stress in the middle of the test bar (the neck, left) as a

function of the local strain, (a), and the resulting engineering stress, which is the tensile force measured right at the clamping side over the initial

cross-section of the test bar, versus engineering strain, defined as the length of the test bar over the original length, see (b), at six different stages

of the test.

10 At the occasion of the 50th anniversary celebration of the

Journal of Polymer Science, the editors selected Henkee and

Kramers original break-through paper of 1984[53]to be reprinted

in 1996. It is still today interesting to read Ed Kramers (and also

Hugh Browns) reflection written on this occasion[54].Please taste

the following sentence: Even today the idea that a linear

viscoelastic measurement of the rubbery plateau modulus in the

melt could elucidate the large strain plastic deformation behavior of

the glass seems nothing short of miraculous; but there were

certainly many hints in previous work that entanglements were

important for both deformation and fracture.



14/24

we counted the number of (radial) crazes that were

caused by indentation as a function of the load applied.

Extrapolating to zero crazes yielded critical normal

forces of ca. 1.5 and 2.5 N, respectively. Performing

numerical simulations of the same tests, using the

advanced constitutive equations developed, we

computed the maximum values of the maximum

hydrostatic stress, somewhere in the sample, during

Fig. 14. Tensile testing of PS. After neck formation at yield strain softening sets in, the stress in the remaining part of the test bar decreasessubstantially due to unloading because of strain localization in the neck. Subsequently, strain hardening occurs in he neck, but it is insufficiently

pronounced. Localization remains in the neck, until fibril failure there occurs.

101 103 105 107 10930

50

70

90(a) (b)

time-to-failure [s]

appliedstress[MPa]

101 102 103 104 10545

50

55

60

65

time-to-failure [s]@1 Hz

appliedmeanstress[MPa]

annealedquenched

annealedquenched

Fig. 15. Life time prediction of PC during creep loading (a) and fatigue loading (b) with symbols the experimental results and full lines the

model predictions without any adjustable parameter[49].



15/24

loading as function of the normal force applied, see

Fig. 17a. For both, the quenched and annealed

samples, this leads to the same critical value for the

hydrostatic stress for PS of 40 MPa, above which a

cavity will form that will lead to the first craze.

Performing the same test for different amorphous

polymers, characterized by their entanglement density

new1/Me, resulted in a semi-log relation giving, e.g.

for PC a critical value of 90 MPa, seeFig. 17b.

Using these critical values, finite element simu-

lations can be performed on notched tensile bars, with

a small defect under the notch representing the fresh

cut of a quenched razor knife, seeFig. 18a.

Similar to the analysis of the micro-indentation

tests, we can compute the maximum hydrostatic

stress, somewhere in the sample, during loading. We

find cavity formation and thus craze initiation in PS

under the defect at low macroscopic strains[61], see

Fig. 18b, while PC survives this problem but exceeds

the critical hydrostatic stress of 90 MPa under the

notch at a macroscopic strain of ca. 1%, see Fig. 18c.

It is because of these problems that polymers need to

be made heterogeneous, e.g. by rubber modification.

What the heterogeneity (soft inclusions into the glassy

matrix) typically should taken care of is to prevent the

critical distance to occur everywhere in the glassy

matrix, such that the tri-axial stress state can not exist,

thus crazing can not start. The typical distance in PS is

very small, seeFig. 18b, and is given by the distance

from the surface to the place where the maximum

hydrostatic stress is found. Although FEM calcu-

lations bear no absolute dimensions, Robert Smit took

experiments as his starting point in modeling and

based thereupon, this distance could be very roughly

estimated using the amplitude of the defect, to be /

1 mm, while for PC the critical distance could be

0 5 10 15 20 25 30

100

150

200

250

300

(a) (b)

(c)(d)

time [s]

temperature[C

]

from surfaceto center

0 0.25 0.5 0.75 1

55

60

65

70

center surface

normalized thickness []

yieldstress[MP

a]

25

30

35

Sa

[]

0 0.25 0.5 0.75 1

50

60

70

center surface

normalized thickness []

yieldstress[MPa]

50C

80C

100C110C120C130C

20 40 60 80 100 120 14055

60

65

70

75

Tg

150C155C

150C155C

mould temperature [C]

yieldstress[MPa]

4mm1mmmodel predictions

25

30

35

40

Sa

[]

Fig. 16. Thermal history within a simple rectangular mould (a) and resulting calculated yield stress (b) and comparison of the averaged

calculated and experimental yield stress as a function of mould temperature used during injection moulding (c and d).



16/24

estimated using the curvature of the notch to be /200 mm[61].

8. Heterogeneous systems

So, although critical distances apparently differ for

different polymers, characterized by the ratio of strain

hardening and strain softening after yield, the need for

heterogeneity is universal and its role is to prevent the

first cavity of the first craze to start to grow. Although

in practice usually rubber is used as impact modifier,we will start the analysis with a zero modulus rubber,

holes, that should prevent the critical triaxial stress state

to occur. The problems met in analyzing heterogeneous

polymer systems were threefold: (i) the complexbehavior of the constituents, (ii) the complex geometry

of the microstructure, and (iii) the unknown relation

between micro events and macroscopic behavior. The

first problem was solved after the development of

reliable 3D quantitative constitutive equations, see

Table 1, the second problem by using FEM, finite

element methods,and for the last problem we developed

the MLFEM, multi-level finite element model[6266].

Basically, MLFEM uses a FEM analysis of the

macroscopic structure and goes one level down in

every integration point of every finite element. The localdisplacements in these integration points are enforced

on the edges of bi-periodic RVEs, representative

volume elements, seeFig. 19.

0 1 2 3 40

10

20

30

40

50

annealed

quenched

load [N]

max.

hydr.stress

[MPa]

100 101 10225

50

75

100(b)(a)

PS

PS/PPE

PC

network density [1025m3]

criticalhydr.stress[MPa]

Fig. 17. Maximum hydrostatic stress in the sample during micro-indentation for quenched and annealed PS (a) and entanglement density

dependence of the critical value of this stress (b).

Fig. 18. Meshed notched tensile bar with defect under the notch (a). Resulting stress profiles during loading, showing that the critical hydrostatic

stress of 40 MPa is reach for PS underneath the defect already at low macroscopic strains of !0.2% (b) while for PC, that survives the defect

sensitivity, the critical hydrostatic stress of 90 MPa is reached below the notch at a macroscopic strain of ca. 1% (c) [61].



17/24

The RVEs are meshed and loaded, after which the

average RVE stress is computed and brought back to

the highest FEM level, to solve the balance equations.

The procedure repeats, until equilibrium is reached.

Then the next incremental strain step is applied. In

this way we use the RVE analysis as a non-closed

form of a constitutive equation for the complex

microstructure. Besides, a local check inside everyRVE on local failure is possible [87]. Fig. 19

compares the RVE responses of PC, Fig. 19a, and

PS, Fig. 19b, discriminating between elastically

loaded and plastically deformed regions. Of course

already at extremely low macroscopic strains,

localizations set in, there where two holes are

accidentally close such that a stress concentration

exists. The material starts to locally flow. In contrast

to PC, that eventually spreads plastic deformation

over the total RVE, in PS it still localizes in bands

perpendicular to the loading direction,Fig. 19b.Reason is the lack of strain hardening in PS, given

its low entanglement density ne or high molecular

weight between entanglementsMe, caused by its large

chain stiffness. As inFig. 14, we again can conclude

via this RVE-analysis that PS is indeed too ductile to

be tough. The thin filaments formed are not able to

transfer load and strain keeps localized until the fibrils

break. The consequences of this RVE response are

clearly found on the macro-level when applying

MLFEM, seeFig. 20.

PC that homogeneously hits the 90 MPa hydro-

static stress criterion at low macroscopic strains,

Fig. 20b, is completely tough in the heterogeneous

case,Fig. 20d. InFig. 20, 30% voids were used, but

also 5% voids prove to be sufficient for PC[65]. This

illustrates that it is quite easy to toughen PC. PS on the

contrary hits in its homogeneous state the 40 MPa

criterion already at very low strains,Fig. 20a, but onlyslightly improves when made heterogeneous,

Fig. 20c. Although clearly improved and comparable

with plain PC, the deformation still localizes in a band

perpendicular to the drawing direction and eventually

the filaments in the RVE will break, see alsoFig. 19b.

9. Optimal toughness modifier

Apparently for too ductile, too easy to flow,

polymers like PS, we need to do a little bit more. Inorder to increase strain hardening, since that is what

we need, we can introduce a smallerMeby mixing PS

with PPE, which is the miscible Norylw blend of GEP

seeFig. 11, or alternatively chemically crosslinked PS

[53].

A more practical way is to let a rubber shell

locally support the straining filaments, not changing

the PS properties, but changing the local structure

properties to make it more strain-hardening, see

Fig. 21. This now seems to be successful. PS is

Fig. 19. Loading representative volume elements, RVEs, of PC (a) and PS (b). Gray is below yield, black represents places that are plastically

deformed[64].



18/24

tough. During loading we did not meet a failure

criterion and it was only since we had to remesh on

the RVE level that we had to stop the computation.

The idea of precavitation of the rubber particles is

based on the fact that we do not want to rely onthe in situ cavitating capabilities of the rubbery

particles during impact loading under low tempera-

tures [6772]. The last two conditions basically

increase the glass transition temperature of the

rubber, making it more and more difficult to cavitate,

which is necessary to prevent the occurrence of too

high tri-axial stresses. (If we let the material self

decide where to cavitate, it will form one localized

craze in the plane perpendicular to the loading

direction, making the polymer brittle.) Besides low

temperatures and high deformation rates, we more-

over want to apply extremely small particles of

the impact modifiers used, despite that they are

difficult to cavitate. Reason is that we than need less

rubber, so modulus and yield stress remain high, but

also that we still can keep optical clarity. An

illustrative example is found in Fig. 22 where for

an in situ copolymerized PMMAaliphatic epoxy

system (8020) with 30 nm dispersed phase, impacttoughness is found, only after precavitating the

samples at low deformation rates[7376].

A second example of precavitating rubber

inclusions can be found in Clive Bucknalls work,

using thermal contraction tests where he supercooled

heterogeneous polymer systems for sufficiently long

times to induce rubber cavitation[77,78]. As a result,

experiments and simulations suggest that the universal

optimal impact modifier is a 3000300303 system

with in a matrix of 3000 MPa (all amorphous

Fig. 20. MLFEM results on loading a notched, scratched tensile bar

of homogeneous and heterogeneous (30% voids) PC (b and d), and

PS (a and c). Strain localizations and delocalizations are shown,

explaining the toughness of these two extreme polymers [65].

Fig. 21. MLFEM results on loading a notched, scratched tensile bar

of homogeneous (a) and heterogeneous PS with 30% voids (b) or

with 30% precavitated rubber particles supporting the filaments (c).

Only in the last case delocalization of strain is found and toughness

is obtained[65].



19/24

polymers) we add a 30 nm dispersion of a precavitated

core-shell rubber with a shell with a pretty high rubber

modulus of 300 MPa and a core that should be a hole,

so with a modulus /3 MPa. We tried to make thesestructures via a CIPS process, chemically induced

phase separation, making use of a self-organizing

process in diblock copolymers [7982]. Nanostruc-

tures were induced and in PS/PBAPEB 75-25 blends

with poly-butylacrylate as a rubber shell and poly-

ethylene-(co)butylene as the liquid core, a change from

crazing to cavitation-induced shear yielding was found

[79,80] while the macroscopic strain of the triblock

system PMMAPBAPEB proved to increase to 140%

[81]. Using a crystallizable polymer as a core, also

cavitation can be obtained as was proven by the systemPSPBPCL, where polybutene was the shell of the

dispersed phase and polycaprolacton the crystallizing

core [82]. In all cases, we still needed a too large

amount of impact modifier. The first one to prove that

an optimal impact modifier indeed can be realized was

Christian Koulic from Jeromes Liege group[8386]

who found that cucumber-type of structures of PS-PIP

in PA, with a size of around 50 nm, gave maximum

toughness with only 4% rubber added, seeFig. 23.

10. Brittle-to-tough transitions

Finally, if we apply the craze initiation criterion

used in Sections 68, on the deformation ofheterogeneous systems, we can find both a

temperature induced brittle-to-tough transition, see

Fig. 24a, as well as a critical interparticle distance

induced transition, see Fig. 24b [87].

The critical interparticle distance originates

from the fact that the Tg of a polymer decreases

at a surface (and increases at a solid interface)

over a depth of 100 nm [8891]. As a conse-

quence, Tg is at room temperature at the surface

and reaches its bulk value only at 100 nm. As a

result, the near-surface mechanical properties aredifferent from that in the bulk [92]. This now

gives an absolute scale in the microstructure, but

also in the finite element calculations, where we

repl ace a ch ang e in Tg by a change in

temperature from the surface into the bulk,

gradually changing over the same 100 nm. The

size of the (constant volume fraction) dispersed

phase now determines whether or not this layer

of lower Tg, thus lower yield stress sy percolates

Fig. 22. In situ X ray experiments on a PMMAaliphatic epoxy system during slow and fast loading [75].



20/24

through the RVE yes or no, thereby determining

toughness.

11. Conclusions

The major conclusions to be drawn are that (i)

toughness is equivalent to delocalization of strain, (ii)

the two types of bonds present make all polymers

intrinsically tough, (iii) polystyrene is too ductile to be

tough (but supporting shells increase strain hardening

of the local structure), (iv) proper constitutive

modeling enables to quantitatively describe polymer

responses, including the effects of temperature, strain

rate, and time (ageing), (v) an independent craze

initiation criterion must be added in order to

numerically predict failure and this criterion can be

experimentally determined by using indentation tests,

(vi) all polymers must be made heterogeneous in order

to circumvent defect or notch sensitivity, (vii)

Fig. 23. Rubber toughened polyamide demonstrating the existence of the optimal impact modifier [8386].

Fig. 24. BTtransitions for heterogeneous PS as a function of temperature (a), and critical interparticle distance (b) [87].



21/24

modeling suggests that the ultimate toughness

modifier is a 3303003000 pre-cavitated core-

shell rubber with core /3 MPa modulus, 30 nm

diameter, shell 300 MPa modulus in a matrix of

3000 MPa, and finally (viii) modeling suggests the

presence of both a temperature induced as well as a

critical interparticle distance induced brittle-to-tough

transition in polymeric materials.

How do the present analyses cope with the

previously empirically determined relation betweenchain stiffness and molecular weight between entan-

glements[93,94]and its relation with drawability and

the transition from crazing to shear yielding[5359],

seeFig. 25.

The answer is straightforward. On the one hand,

the strain hardening modulus proves to be directly

related toMe, while a larger modulus gives less strain

softening thus less localization changing the defor-

mation behavior from crazing to shear yielding, see

Figs. 1114.On the other hand, there exists a relation

between the critical stress for craze initiation andMe,seeFig. 17b, making the higher entangled polymers

less sensitive for scratches and easier to rubber

modify, seeFigs. 1821.

Acknowledgements

This paper is based on a number of successive and

ongoing PhD studies in our group: Marco van der

Sanden (1993) on the concept of ultimate toughness,

Theo Tervoort (1996) on constitutive modeling, Peter

Timmermans (1997) on modeling of necking, Robert

Smit (1998) on the multi-level finite element method,

Bernd Jansen (1998) on microstructures for ultimate

toughness, Harold van Melick (2002) on quantitative

modeling, Bernard Schrauwen (2003) on semi-

crystalline polymers, Hans van Dommelen (2003)

on multi-scale semi-crystalline, Ilse van Casteren

(2003) on nanostructures for ultimate toughness,Edwin Klompen (2005) on long-term prediction,

Jules Kierkels (2006) on toughness in films, Roel

Janssen (2006) on creep rupture and fatigue, and Tom

Engels (2007) on coupling of processing with proper-

ties. Theses are downloadable from (please select by

year) http://www.mate.tue.nl/mate/publications/

index.php/4. The research was sponsored by the

Dutch Polymer Insitute, DPI, the national science

foundation, STW, and the university, TU/e.

References

[1] GSell C, Jonas JJ. Yield and transient effects during the

plastic deformation of solid polymers. J Mater Sci 1981;16(7):

195674.

[2] GSell C, Hiver JM, Dahoun A, Souahi A. Video-controlled

tensile testing of polymers and metals beyond the necking

point. J Mater Sci 1992;27(18):50319.

[3] Arruda EM, Boyce MC. Evolution of plastic anisotropy in

amorphous polymers during finite straining. Int J Plast 1993;

9(6):697720.

Fig. 25. Extension ratio in crazes and deformation zones (DZs) as a function of network density n (w1/Me) in (un)crosslinked PS films (a).

Similar plot for different polymers (PPO is the somewhat confusing old notation for poly-phenylene-ether, PPE) (b). Reproduced withpermission from[53].

http://www.mate.tue.nl/mate/publications/index.php/4http://www.mate.tue.nl/mate/publications/index.php/4http://www.mate.tue.nl/mate/publications/index.php/4http://www.mate.tue.nl/mate/publications/index.php/4


22/24

[4] Boyce MC, Arruda EM, Jayachandran R. The large strain

compression, tension, and simple shear of polycarbonate.

Polym Eng Sci 1994;34(9):71625.

[5] McLeish TCB, Milner ST. Entangled dynamics and melt flow

of branched polymers. Adv Polym Sci 1999;143(BranchedPolymers II):195256.

[6] Read DJ, McLeish TCB. Molecular rheology and statistics of

long chain branched metallocene-catalyzed polyolefins.

Macromolecules 2001;34(6):192845.

[7] Swartjes FHM, Peters GWM, Rastogi S, Meijer HEH. Stress

induced crystallization in elongational flow. Int Polym Proc

2003;18(1):5366.

[8] Tervoort TA, Visjager JF, Smith P. Melt-processable

poly(tetrafluoroethylene)-compounding, fillers and dyes.

J Fluorine Chem 2002;114(2):1337.

[9] Tervoort TA, Visjager JF, Graf B, Smith P. Melt-processable

poly(tetrafluoroethylene). Macromolecules 2000;33(17):

64605.

[10] Tervoort TA, Visjager JF, Smith P. On abrasive wear of

polyethylene. Macromolecules 2002;35(22):846771.

[11] Smith P, Lemstra PJ, Kalb B, Pennings AJ. Ultrahigh-strength

polyethylene filaments by solution spinning and hot drawing.

Polym Bull 1979;1(11):7336.

[12] Smith P, Lemstra PJ. Ultrahigh-strength polyethylene fila-

ments by solution spinning/drawing. 2. Influence of solvent on

the drawability. Makromolekulare Chemie 1979;180(12):

29836.

[13] Smith P, Lemstra PJ. Ultrahigh strength polyethylene

filaments by solution spinning/drawing. 3. Influence of

drawing temperature. Polymer 1980;21(11):13413.

[14] Kwolek SL, US Patent 3671 542; 1972.

[15] Yang HH. Aromatic high-strength fibers. New York: Wiley;1989.

[16] Gordon JE. Structures, or why things dont fall down.: Da

Capo Press; 1978 p. 74.

[17] Clyne TW, Phillipps AJ. Interfacial control and macroscopic

failure in long-fiber-reinforced and laminated inorganic

composites. Comp Sci Tech 1994;51(2):27182.

[18] Smith BL, Schaffer TE, Viani M, Thompson JB,

Frederick NA, Kind J, et al. Molecular mechanistic origin of

the toughness of natural adhesives, fibers and composites.

Nature 1999;399(6738):7613.

[19] Wang RZ, Suo Z, Evans AG, Yao N, Aksay IA. Deformation

mechanisms in nacre. J Mater Res 2001;16(9):248593.

[20] Haward RN, Thackray G. Use of a mathematical model to

describe isothermal stress-strain curves in glassy thermo-

plastics. Proc R Soc London Ser A 1967;302(1471):45372.

[21] Govaert LE. Bob Haward, a chemist on the quest for strength.

J Polym Sci Part B 2004;42(11):iiiiiv.

[22] Boyce MC, Parks DM, Argon AS. Mech Mater 1988;7:

1533.

[23] Hasan OA, Boyce MC. A constitutive model for the nonlinear

viscoelastic viscoplastic behaviour of glassy polymers. Polym

Eng Sci 1995;35:33144.

[24] Buckley CP, Jones DC. Glass-rubber constitutive model for

amorphous polymers near the glass transition. Polymer 1995;

36:330112.

[25] Dooling PJ, Buckley CP, Hinduja S. The onset of nonlinear

viscoelasticity in multiaxial creep of glassy polymers: a

constitutive model and its application to PMMA. Polym Eng

Sci 1998;38:892904.

[26] Gerlach C, Buckley CP, Jones DP. Development of anintegrated approach to modelling of polymer film orientation

processes. Trans Inst Chem Eng Part A 1998;76:3844.

[27] Tervoort TA, Klompen ETJ, Govaert LE. A multi-mode

approach to finite, three-dimensional, nonlinear viscoe-

lastic behaviour of polymer glasses. J Rheol 1996;40:

77997.

[28] Govaert LE, Timmermans PHM, Brekelmans WAM. The

influence of intrinsic strain softening on strain localisation in

polycarbonate: modeling and experimental validation. J Eng

Mater Technol 2000;122:17785.

[29] Klompen ETJ, Engels TAP, Govaert LE, Meijer HEH.

Modelling of the post-yield response of glassy polymers:

influence of thermomechanical history. Macromolecules 2005,38(16):69977008.

[30] Loo LS, Cohen RE, Gleason KK. Chain mobility in the

amorphous region of nylon 6 observed under active uniaxial

deformation. Science 2000;288(5463):1169.

[31] Capaldi FM, Boyce MC, Rutledge GC. Enhanced mobility

accompanies the active deformation of a glassy amorphous

polymer. Phys Rev Lett 2002;89(17):175505.

[32] Theodorou DN, Suter UW. Local structure and the mechanism

of response to elastic deformation in a glassy polymer.


[33] van Melick HGH, Govaert LE, Meijer HEH. Localization

phenomena in glassy polymers: influence of thermal and

mechanical history. Polymer 2003;44(12):357991.[34] McKenna GB. Mechanical rejuvenation in polymer glasses:

fact or fallacy? J Phys Cond Matter 2003;15(11):S737S63.

[35] Tervoort TA, Govaert LE. Strain-hardening behavior of

polycarbonate in the glassy state. J Rheol 2000;44(6):

126377.

[36] Govaert LE, van Melick HGH, Meijer HEH. Temporary

toughening of polystyrene through mechanical pre-condition-

ing. Polymer 2001;42(3):12714.

[37] van Melick HGH, Govaert LE, Raas B, Nauta WJ,

Meijer HEH. Kinetics of ageing and re-embrittlement of

mechanically rejuvenated polystyrene. Polymer 2003;44(4):

11719.

[38] Struik LCE. Private communications.

[39] Hodge IM. Enthalpy relaxation and recovery in amorphous

materials. J Non-Cryst Solid 1994;169:21166.

[40] Hutchinson JM. Physical aging of polymers. Prog Polym Sci

1995;20:70360.

[41] Struik LCE. Physical ageing of amorphous polymers and other

materials. Amsterdam: Elsevier; 1978.

[42] Aref-Azar A, Biddlestone F, Hay JN, Haward RN. The effect

of physical ageing on the properties of poly(ethylene

terephthalate). Polymer 1983;24(10):124551.

[43] Utz M, Debenedetti PG, Stillinger FH. Atomistic simulation of

aging and rejuvenation in glasses. Phys Rev Lett 2000;84(7):

14714.



23/24

[44] Nandagopal M, Utz M. Thermal versus deformation-induced

relaxation in a glass-forming fluid. J Chem Phys 2003;

118(18):83737.

[45] Lei Y, Cummins K, Lacks DJ. First-principles enthalpy

landscape analysis of structural recovery in glasses. J PolymSci, Part B 2003;41(19):23026.

[46] Lacks DJ, Osborne MJ. Energy landscape picture of overaging

and rejuvenation in a sheared glass. Phys Rev Lett 2004;

93(25):255501/1255501/4.

[47] Wu JJ, Buckley CP. Plastic deformation of glassy polystyrene:

A unified model of yield and the role of chain length. J Polym

Sci Part B 2004;42(11):202740.

[48] van Melick HGH, Govaert LE, Meijer HEH. On the origin of

strain hardening in glassy polymers. Polymer 2003;44(8):

2493502.

[49] Klompen ETJ, Govaert LE, Engels TAP, van Breemen LCA,

Schreurs PJG, Meijer HEH. Quantitative prediction of long-

term failure in polycarbonate, Macromolecules 2005;38(16):

700917.

[50] Govaert LE, Engels TAP, Klompen ETJ, Peters GWM, Meijer

HEH. Processing induced properties of glassy polymers: yield

stress development in polycarbonate. Int Polym Proc 2005;

20(2):1707.

[51] Kramer EJ. Microscopic and molecular fundamentals of

crazing. Adv Polym Sci 1983;5253:156.

[52] Kramer EJ, Berger LL. Fundamental processes of craze

growth and fracture. Adv Polym Sci 1990;91/92:168

[Crazing Polym., Vol. 2].

[53] Henkee CS, Kramer EJ. Crazing and shear deformation in

crosslinked polystyrene. J Polym Sci, Polym Phys Ed 1996;

22(4):72137. Reprinted from J Polym Sci, Part B

1996;34(17) 282541.[54] Kramer EJ. Reflections on Crazing and shear deformation in

crosslinked polystyrene, by Chris S. Henkee, Edward J.

Kramer. J Polym Sci, Polym Phys Ed, 1984;22:721. Comment.

J Polym Sci, Part B 1996:34(17) 282122.

[55] Donald AM, Kramer EJ. The mechanism for craze-tip advance

in glassy polymers. Philos Mag A 1981;43(4):85770.

[56] Donald AM, Kramer EJ. Plane stress deformation zones

at crack tips in polycarbonate. J Mater Sci 1981;16(11):

296776.

[57] Donald AM, Kramer EJ. Effect of molecular entanglements on

craze microstructure in glassy polymers. J Polym Sci, Polym

Phys Ed 1982;20(5):899909.

[58] Donald AM, Kramer EJ. Deformation zones and entangle-

ments in glassy polymers. Polymer 1982;23(8):11838.

[59] Donald AM, Kramer EJ, Kambour RP. Interaction of crazes

with preexisting shear bands in glassy polymers. J Mater Sci

1982;17(6):173944.

[60] Van Melick HGH, Bressers OFJT, Den Toonder JMJ,

Govaert LE, Meijer HEH. A micro-indentation method for

probing the craze-initiation stress in glassy polymers. Polymer

2003;44(8):248191.

[61] Smit RJM, Brekelmans WAM, Meijer HEH. Predictive

modelling of the properties and toughness of polymeric

materials. Part I. Why is polystyrene brittle and polycarbonate

tough? J Mater Sci 2000;35(11):285567.

[62] Smit RJM, Brekelmans WAM, Meijer HEH. Prediction of the

mechanical behavior of nonlinear heterogeneous systems by

multilevel finite element modeling. Comput Methods Appl

Mech Eng 1998;155:18192.

[63] Smit RJM, Brekelmans WAM, Meijer HEH. Prediction of thelarge-strain mechanical response of heterogeneous polymer

systems: local and global deformation behavior of a

representative volume element of voided polycarbonate.

J Mech Phys Solid 1999;47(2):20121.



materials. Part II. Effect of microstructural properties on the

macroscopic response of rubber-modified polymers. J Mater

Sci 2000;35(11):286979.



materials part III simultaneous prediction of micro- and

macrostructural deformation of rubber-modified polymers.

J Mater Sci 2000;35(11):288192.

[66] Meijer HEH, Govaert LE, Smit RJM. A multi-level finite

element method for modeling rubber-toughened amorphous

polymers. ACS Symp Ser 2000;759(Toughening of Plastics):

5070.

[67] Bucknall CB, Karpodinis A, Zhang XC. A model for particle

cavitation in rubber-toughened plastics. J Mater Sci 1994;

29(13):337783.

[68] Ayre DS, Bucknall CB. Particle cavitation in rubber-

toughened PMMA: experimental testing of the energy-balance

criterion. Polymer 1998;39(20):478591.

[69] Lin CS, Ayre DS, Bucknall CB. A dynamic mechanical

technique for detecting rubber particle cavitation in toughened

plastics. J Mater Sci Lett 1998;17(8):66971.[70] Bucknall CB, Rizzieri R, Moore DR. Detection of incipient

rubber particle cavitation in toughened PMMA using dynamic

mechanical tests. Polymer 2000;41(11):414956.

[71] Lazzeri A, Bucknall CB. Recent developments in the

modeling of dilatational yielding in toughened plastics. ACS

Symp Ser 2000;759(Toughening of Plastics):1435.

[72] Bucknall CB. Blends containing core-shell impact modifiers

part 1. Structure and tensile deformation mechanisms. Pure

Appl Chem 2001;73(6):897912.

[73] Jansen BJP, Rastogi S, Meijer HEH, Lemstra PJ. Rubber-

modified glassy amorphous polymers prepared via chemically

induced phase separation. 1. Morphology development and

mechanical properties. Macromolecules 2001;34(12):

39984006.


modified glassy amorphous polymers prepared via

chemically induced phase separation. 2. Mode of microscopic

deformation studied by in-situ small-angle X-ray scattering

during tensile deformation. Macromolecules 2001;34(12):

400718.


modified glassy amorphous polymers prepared via chemically

induced phase separation. 3. Influence of the strain rate on the

microscopic deformation mechanism. Macromolecules 1999;

32(19):62839.



24/24


modified glassy amorphous polymers prepared via

chemically induced phase separation. 4. Comparison of

properties of semi- and full-IPNs, and copolymers of

acrylate-aliphatic epoxy systems. Macromolecules 1999;32(19):62907.

[77] Bucknall CB, Ayre DS, Dijkstra DJ. Detection of rubber

particle cavitation in toughened plastics using thermal

contraction tests. Polymer 2000;41(15):593747.

[78] Bucknall CB, Rizzieri R, Moore DR. Detection of cavitation in

rubber phase of rubber toughened poly(methyl methacrylate)

using thermal contraction measurements. Plast Rub Comp

2001;30(4):1839.

[79] Van Casteren IA, Van Trier RAM, Goossens JGP,

Meijer HEH, Lemstra PJ. The influence of hydrogen

bonding on the preparation and mechanical properties of PS-

diblock copolymer blends. J Polym Sci, Part B 2004;42(11):

213760.

[80] van Asselen OLJ, van Casteren IA, Goossens JGP,

Meijer HEH. Deformation behavior of triblock copoly-

mers based on polystyrene: an FT-IR spectroscopy

study. Macromol Symp 2004;205(Polymer Spec-

troscopy):8594.

[81] Van Casteren IA. Control of microstructures to induce

ductility in brittle amorphous polymers. Dissertation Eindho-

ven University of Technology; 2003. p. 10314 [chapter 6]

(download athttp://www.mate.tue.nl/mate/pdfs/3220.pdf).

[82] Van Casteren IA. Control of microstructures to induce

ductility in brittle amorphous polymers. Dissertation Eindho-

ven University of Technology; 2003. p. 11538 [chapter 7]

(download athttp://www.mate.tue.nl/mate/pdfs/3220.pdf).

[83] Koulic C, Jerome R. Nanostructure PMMA: from lamellarsheets to double-layered vesicles. Macromolecules 2004;

37(3):88893.

[84] Koulic C, Jerome R. Nanostructured polyamide by reactive

blending. 1. Effect of the reactive diblock composition.


[85] Koulic C, Jerome R. Nanostructured polyamide by reactive

blending: 2. Transition from nanovesicles to cucumber-like

core-shell nano objects. Prog Colloid Polym Sci 2004;129:

705.

[86] Koulic C. Reactive blending as a tool towards nanostructuredpolymers. Dissertation University of Liege; 2004. p. 8596

[chapter 5].

[87] van Melick HGH, Govaert LE, Meijer HEH. Prediction of

brittle-to-ductile transitions in polystyrene. Polymer 2003;

44(2):45765.

[88] Forrest JA, Mattsson j. Reductions of the glass transition

temperature in thin polymer films: probing the length

scale of cooperative dynamics. Phys Rev E 2000;61(1):

R53R6.

[89] Long D, Sotta P. How the shift of the glass transition

temperature of thin polymer films depends on the adsorption

with the substrate Los Alamos National Laboratory, Preprint

Archive, Condensed Matter 2003:arXiv:cond-mat/0301100

2003 p. 133.

[90] Sotta P, Long D. The crossover from 2D to 3D percolation and

its relationship to glass transition in thin films. Theory and

numerical simulations Los Alamos National Laboratory,

Preprint Archive, Condensed Matter 2003:arXiv:cond-

mat/0301120 2003 p. 115.

[91] van Zanten JH, Wallace WE, Wu W-L. Effect of strongly

favorable substrate interactions on the thermal properties

of ultrathin polymer films. Phys Rev E 1996;53(3):

R2053R6.

[92] van Melick HGH, van Dijken A, den Toonder JMJ,

Govaert LE, Meijer HEH. Near-surface mechanical properties

of amorphous polymers. Philos Mag A 2002;82(10):

2093102.[93] Wu S. Chain structure and entanglement. J Polym Sci, Part B

1989;27(4):72341.

[94] Wu S. Chain structure, phase morphology, and toughness

relationships in polymers and blends. Polym Eng Sci 1990;

30(13):75361.

http://www.mate.tue.nl/mate/pdfs/3220.pdfhttp://www.mate.tue.nl/mate/pdfs/3220.pdfhttp://www.mate.tue.nl/mate/pdfs/3220.pdfhttp://www.mate.tue.nl/mate/pdfs/3220.pdf

article relationships material properties

Documents